GPU Performance Engineer - Neural Reconstruction

Nvidia

Remote Actively hiring Posted this week
Remote (Us, Ca, Remote, US) Posted 6 days ago $224,000–$356,500 / year
View original post Log in to save

At a glance

AI generated

TL;DR

Join NVIDIA as a GPU Performance Engineer for Neural Reconstruction, a role that demands expertise in optimizing 3D reconstruction workflows using cutting-edge technologies like Gaussian Splatting. You will profile and enhance CUDA and PyTorch performance across various stages of neural reconstruction, from data loading to rendering, ensuring efficient memory usage and multi-GPU execution. Utilizing tools such as Nsight Systems and Compute, you’ll identify bottlenecks and develop benchmarks to maintain quality and reliability in production systems. Ideal candidates have extensive experience with Python, C++, PyTorch, and CUDA, along with a deep understanding of GPU performance optimization techniques and 3D vision goals. This role requires collaboration across research and engineering teams to deliver robust, scalable solutions for complex ML workloads.

Skills

Python C++ PyTorch CUDA Nsight Systems Nsight Compute NVTX PyTorch Profiler Gaussian Splatting NeRF differentiable rendering rasterization neural rendering SLAM 3D reconstruction camera and lidar geometry projection models calibration rolling shutter depth rendering multi-sensor reconstruction distributed training memory footprint optimization

What you'll do

  • Profile end-to-end neural reconstruction workflows to identify performance bottlenecks.
  • Improve CUDA and PyTorch performance for Gaussian Splatting and related workloads.
  • Analyze GPU performance using Nsight Systems, Compute, NVTX, and other profiling tools.
  • Optimize sparse rendering workloads including tile-level masking/culling and multi-GPU execution.
  • Translate Python/PyTorch bottlenecks into efficient CUDA/C++ implementations when necessary.

What we're looking for

  • 12+ years of experience in Computer Science or related field with strong programming skills in Python and C++.
  • Hands-on experience optimizing GPU-accelerated workloads using CUDA and PyTorch.
  • Proficiency in profiling and performance analysis for CPU/GPU bottlenecks and memory pressure.
  • Ability to develop benchmarks and validate optimizations while preserving correctness and quality.
  • Experience with Gaussian Splatting, NeRF, differentiable rendering, or 3D reconstruction pipelines.

Employer

About Nvidia

Nvidia is a leading designer of graphics processing units (GPUs) and system-on-chip units, powering gaming, professional visualization, data centers, and artificial intelligence workloads. Industry: Semiconductors & AI Computing

Nvidia currently has 825 open roles on FindRole.

Listed pay typically runs $184,000–$287,500 across 813 roles with salary data.

Most-posted roles

View all roles at Nvidia